Closed Happyhappyhappy23 closed 1 year ago
Sorry, I don't fully get your question. The screenshot in the paper that you attached is exactly what we did. The corresponding code is in this function https://github.com/fjxmlzn/DoppelGANger/blob/1e16b4e023f5ed4e5b82e878d4896d40eaf097b9/gan/network.py#L170
Is there actually only one MLP in the code? "Real" and "fake" metadata are generated at the same time?